CN1262986C

CN1262986C - Image processing method and image processing apparatus

Info

Publication number: CN1262986C
Application number: CNB018113346A
Authority: CN
Inventors: 大喜智明
Original assignee: Matsushita Electric Industrial Co Ltd
Current assignee: Panasonic Holdings Corp
Priority date: 2000-04-21
Filing date: 2001-04-20
Publication date: 2006-07-05
Anticipated expiration: 2021-04-20
Also published as: US20030099410A1; JP4880172B2; KR20020089517A; CN1441944A; KR100820507B1; WO2001082286A1; US7116842B2; TWI229842B

Abstract

The object of the present invention is to provide an image processing method and an image processing device, which can obtain a high-quality image without spots and appear harmonious when the number of pixels of a character image including thin lines is converted. Therefore, in the input original image signal, the number of pixels constituting the line segment is converted, and when the number of pixels is converted in at least one direction of the row direction and the column direction of the matrix at a specified pixel number conversion rate, The converted line constituting pixels corresponding to the line constituting pixels in the input original image signal are adjusted to be continuous in a certain number determined based on the conversion rate of the number of pixels.

Description

Image processing method and image-processing system

Technical field

The present invention relates to import former image, it is transformed into the image processing method and the image-processing system of the image of different number of pixels.

Background technology

In personal computer (below, be called computer) and TV receiver etc., transmit image when carrying out the demonstration of picture from eikongen to display, the situation that useful image-processing system carries out the number of pixels conversion with the interconversion rate of expectation.

The representative of the number of pixels converter technique that adopts as this image-processing system, the known brightness value that a plurality of pixels around the pixel that adopts before the conversion are arranged carries out the Cubic interpolation method and the linear interpolation of interpolation, and former state adopts the nearest interpolation method etc. of the gray-scale value of the pixel in the nearest coordinate position of pixel coordinates before conversion unchangeably.

, the situation that shows natural image and character and picture is simultaneously arranged in computer and TV, in the image of arbitrary type, carry out clearly to show after the number of pixels conversion.

Here, according to above-mentioned Cubic interpolation method and linear interpolation, though remain unchanged before and after the centre of gravity place conversion of gray scale, can obtain to keep natural image original appearance and the few good conversion image of whole visual spot for natural image, but but be not suitable for character and picture (literal and figure etc. have the image of sharp-pointed marginal portion), thicken in the marginal portion.

On the other hand, recently interpolation method is owing to lose the natural image original appearance easily when conversion, thereby is not suitable for the conversion natural image; And when the conversion character and picture, the marginal portion after the conversion can not thicken.But, if adopt nearest interpolation method that character and picture is carried out the number of pixels conversion, the unbalanced problem of live width then takes place, that is, the live width before the conversion is identical, along with the difference of the position of line causes the different problem of live width after the conversion.

Particularly in the image of handling with computer, the literal of former image and the live width of chart often are made of the fine rule section of 1 pixel, if carry out the number of pixels conversion with nearest interpolation method, then the live width inequality in the image after the conversion seems very eye-catching, gives the sensation of not harmony.

In addition, select above-mentioned Cubic interpolation method and recently interpolation method to carry out the image-processing system of demonstration of natural image and character and picture also just under development.

For example, the spy opens in the flat 11-203467 communique, as shown in figure 23, the image-processing system that is made of waveform recognition circuit 31, selector switch 32, Cubic interpolation circuit 33 and nearest interpolation circuit 34 has been described.

In this device, input image signal 35 is by 31 identifications of waveform recognition circuit, selector switch 32 is generally selected Cubic interpolation circuit 33 according to this result, and selects nearest interpolation circuit 34 when being identified as staircase waveform in the marginal portion, thereby, natural image is adopted the Cubic interpolation method, can obtain to be suitable for the pixel images of natural image, simultaneously, image to the marginal portion adopts nearest interpolation method, can obtain to have reduced the image of spot.

But, produce the problem of live width inequality in the character and picture in this case equally.

With reference to Figure 24 the problems referred to above are described.

Among Figure 24, (a), (c) carry out the situation of interpolation for adopting nearest interpolation method, (b) in the horizontal direction for adopting the cubic interpolation method.Represent among the figure that input imagery is made of horizontal direction 5 pixel coordinates H1～H5 arranged side by side, then, this input imagery is transformed into the situation of pixel coordinates D1 arranged side by side in the horizontal direction～D6....

Among Figure 24, white pixel is represented high brightness (brightness value 255), and black pixel is represented low-light level (brightness value 0), low-light level during the diagonal line hatches pixel is represented (brightness value 64), and the stain shadow pixels is represented intermediate luminance (brightness value 128).

Figure 24 (a) (b) in, the H3 in the input imagery is low-light level, other pixel H1, H2, H4, H5 are high brightness, at this moment, H3 is for constituting the pixel of line.In addition, among Figure 24 (c), the H4 in the input imagery is for constituting the pixel of line.

At first, as Figure 24 (b), when carrying out the Cubic interpolation, each pixel D2, the D3 after the conversion, D4, D5 carry out interpolation (for example the coordinate D 3 of output terminal carries out interpolation with reference to the coordinate H1～H4 of input imagery end) with reference to 4 pixels before the conversion respectively.

At this moment, in the image of input, (D2～D5) reduce with the luminance difference of pixel (D1, D6) on every side simultaneously, thereby line thickens because a pixel (H3) wide increases to 4 pixels after conversion.

Then, at Figure 24 (a), (c) all carry out nearest interpolation in, shown in (a), H3 is that line segment constitutes under the situation of pixel, and D3, D4 become line segment formation pixel in output terminal, in addition, shown in (c), H4 is that line segment constitutes under the situation of pixel, has only D5 to become line segment in output terminal and constitutes pixel.Like this, the image that live width is all 1 pixel in input end with nearest interpolation method conversion after, it is wide to become 2 pixels in (a), it is wide to become 1 pixel in (c).

Summary of the invention

The object of the present invention is to provide image processing method and image-processing system, when the character and picture that comprises fine rule carries out the number of pixels conversion, can obtain the excellent picture of the immaculate and the harmony that seems.

Thereby, input matrix of the present invention shows the former image of usefulness, with predetermined number of pixels interconversion rate when at least one assigned direction carries out the number of pixels conversion in line direction and the column direction, obtain the relevant characteristic information of continuous number of the line segment formation pixel the former image with input from the direction of this appointment, according to the characteristic information that obtains, with a certain number of determining based on the number of pixels interconversion rate, will constitute pixel with the line segment after line segment constitutes the corresponding conversion of pixel and be adjusted into continuously.

Above-mentioned " line segment formation pixel " is the pixel that constitutes line segment in literal and the ruling etc.

Here, the position that constitutes pixel of the line segment after the conversion line segment of preferably adjusting to input substantially as far as possible constitute pixel near.

The occasion that the nearest interpolation method of the employing that illustrates in conventional art is carried out the number of pixels conversion, as above-mentioned, the live width difference after the conversion because the live width of preceding literal of conversion and ruling is identical, thereby the not harmony that seems, and the invention described above is when the number of pixels conversion, (especially live width is 1 pixel by line segments such as literal in the preferential maintenance input imagery and rulings, the fine rule of 2 pixels) live width is carried out the conversion of number of pixels, compare with the situation of carrying out the number of pixels conversion with nearest interpolation method, though the position of line segment more or less is offset, but the live width after the conversion keeps homogeneous, thereby can significantly reduce not harmony sense.

Like this, the homogeneity by preferential maintenance live width can reduce not harmony sense.This be because, in general, for the line segment (especially live width is the fine rule of 1 pixel, 2 pixels) of literal in the image and ruling, with the position of line segment whether have the skew compare, whether live width keeps the influence of all not harmony senses of a pair of image bigger.

As the concrete method of in the homogeneous of the above-mentioned live width of maintenance, carrying out the number of pixels conversion, can enumerate following example.

Method one detects the line segment that the former image of input comprises and constitutes pixel, with a certain number of determining based on the number of pixels interconversion rate, the line segment that detects is constituted line segment after the conversion of pixel correspondence and constitutes pixel and be adjusted on the direction of appointment continuously.

Method two detects the edge that the former image of input comprises and constitutes pixel, constitutes number of pixels with the edge that detects and multiply by based on the number behind the well-determined integer of number of pixels interconversion rate, the edge after the conversion is constituted pixel be adjusted on the direction of appointment continuously.

Method three is by the number of pixels transform method of the 1st interpolation method, the combination of the 2nd interpolation method.Wherein, the 1st interpolation method, with a certain number of determining based on the number of pixels interconversion rate, the pixel behind near the output transform input pixel; The 2nd interpolation method, the pattern and the predetermined map table that constitute pixel by the line segment that comprises of former image to input carry out pattern match, the pixel after the output conversion corresponding with importing pixel.

Among the present invention, whether the image of also judging input has the feature of character and picture, in characteristic occasion, adopt the above-mentioned preferential interpolation method of live width to carry out conversion, in the occasion of the feature that does not have character and picture, adopt the center of gravity that is called linear interpolation or Cubic interpolation method to keep interpolation method to carry out conversion.

Thereby, for natural image, can keep the conversion of natural image original appearance, simultaneously,, can keep the conversion of live width homogeneous for character and picture.

In addition, when the number of pixels interconversion rate is similar to the occasion of the multiple of integer, also can adopt the preferential interpolation method of above-mentioned live width to carry out conversion.

A kind of image processing method of the present invention, input matrix shows the former image of usefulness ,-with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it may further comprise the steps:

The edge detects step, detects the edge that comprises in the former image of input and constitutes pixel;

The output set-up procedure detects edge that step detects with described edge and constitutes the continuous number of pixel and multiply by based on the number behind the well-determined integer of number of pixels interconversion rate, the edge after the conversion is constituted pixel be adjusted on the direction of described appointment continuously.

Another image processing method of the present invention, input matrix shows the former image of usefulness, with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it is characterized in that, it selects one of the 1st interpolation method and the 2nd interpolation method to carry out the number of pixels conversion, wherein:

The 1st interpolation method with a certain multiplying power of determining based on the number of pixels interconversion rate, is exported the image after the conversion corresponding with input imagery;

The 2nd interpolation method, the pattern and the predetermined map table that constitute pixel by the line segment that comprises of former image to input carry out pattern match, the pixel after the output conversion corresponding with importing pixel.

Another image processing method of the present invention, input matrix shows the former image of usefulness, with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it is characterized in that comprising:

The 1st interpolation step with a certain multiplying power of determining based on the number of pixels interconversion rate, is exported the image after the conversion corresponding with input imagery;

The 2nd interpolation step, the pattern and the predetermined map table that constitute pixel by the line segment that comprises of former image to input carry out pattern match, the pixel after the output conversion corresponding with importing pixel;

Select step, select the output of one of described the 1st interpolation step and the 2nd interpolation step.

Another image processing method of the present invention, select one of following method to carry out the number of pixels conversion:

The first number of pixels transform method makes the line segment that possesses same live width in the former image of input also keep same live width after conversion;

The second number of pixels transform method remains unchanged the centre of gravity place of conversion front and back,

Comprise:

The feature determination step carries out the whether judgement of characteristic one by one to each zone of former image;

Select step, be judged to be the image area of characteristic, select the described first number of pixels transform method,, select the described second number of pixels transform method for the image area that is judged to be non-characteristic for described feature determination step,

In the described feature determination step,

The line segment that includes in the certain zone that in the input imagery with the target pixel is the center below the regulation live width constitutes pixel, and the coefficient of alteration of the brightness value of group of pixels was respectively below setting around the brightness value of line segment formation group of pixels reached, the mean value of the gray-scale value of described line segment formation group of pixels and the difference of the described mean value of the gray-scale value of group of pixels on every side are judged to be characteristic when setting is above.

Comprise:

In the described selection step,

Whether the value of judging the number of pixels interconversion rate near integer,

Judgement is selected the first number of pixels transform method during near integer, judges when keeping off integer, selects the second number of pixels transform method.

Comprise:

Select step, be judged to be the image area of characteristic, select the described first number of pixels transform method,, select the described second number of pixels transform method for the image area that is judged to be non-characteristic for described feature determination step, in the described selection step,

The difference of the value of number of pixels interconversion rate and approaching with it integer is selected the first number of pixels transform method during less than setting, during greater than setting, selects the second number of pixels transform method.

A kind of image-processing system of the present invention, input matrix shows the former image of usefulness, with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it comprises with the lower part:

The edge detects part, detects the edge that comprises in the former image of input and constitutes pixel;

The output adjustment member, the edge that detects for edge detecting element branch constitutes pixel, on the direction of described appointment,, the edge after the conversion is constituted pixel be adjusted into continuously multiply by a certain number that obtains after the number of pixels based on the well-determined integer of number of pixels interconversion rate.

Another image-processing system of the present invention, input matrix shows the former image of usefulness, with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it is characterized in that, it selects one of the 1st interpolation method and the 2nd interpolation method to carry out the number of pixels conversion, wherein:

Pattern and predetermined map table that the 2nd interpolation method, the line segment that comprises by the former image that will import constitute pixel carry out pattern match, the pixel after the output conversion corresponding with importing pixel.

Another image-processing system of the present invention, input matrix shows the former image of usefulness, with predetermined number of pixels interconversion rate at least one assigned direction in line direction and the column direction is carried out the number of pixels conversion, it is characterized in that comprising:

The 1st interpolation part with a certain multiplying power of determining based on the number of pixels interconversion rate, is exported the image after the conversion corresponding with input imagery;

Pattern and predetermined map table that the 2nd interpolation part, the line segment that comprises by the former image that will import constitute pixel carry out pattern match, the pixel after the output conversion corresponding with importing pixel;

Select part, select the output of one of described the 1st interpolation part and the 2nd interpolation part.

Another image-processing system of the present invention, select one of following method to carry out the number of pixels conversion:

Comprise:

Feature is judged part, and the whether judgement of characteristic is carried out in each zone of former image one by one;

Select part, for the image area that is judged to be characteristic in the described feature determination step, select the described first number of pixels transform method, judge that for described feature part is judged to be the image area of non-characteristic, select the described second number of pixels transform method

Described feature is judged part,

Comprise:

The live width of described regulation is 1 pixel.

Comprise:

In the described selection part,

Comprise:

In the described selection part,

The simple declaration of drawing

Fig. 1 is the block scheme of the image-processing system of embodiment 1.

Fig. 2 is that the characteristic of embodiment 1 detects and line segment detects the process flow diagram of operation.

Fig. 3 is the figure of the input image signal of an example when being in certain area coverage.

Fig. 4 is the process flow diagram that the live width of embodiment 1 is calculated.

Fig. 5 is the process flow diagram of operation of the preferential interpolating circuit of live width of embodiment 1.

Fig. 6 is the process flow diagram of operation of the selection circuit of embodiment 1.

Fig. 7 is the comparison diagram of the preferential interpolation method of live width and other interpolating methods.

Fig. 8 is the block scheme of the image-processing system of embodiment 2.

Fig. 9 is the process flow diagram of operation of the selection circuit of image-processing system.

Figure 10 is the block scheme of the image-processing system of embodiment 3.

Figure 11 is the process flow diagram that the continuous edge number of embodiment 3 calculates.

Figure 12 is the process flow diagram of an example of operation of the preferential interpolating circuit of live width of embodiment 3.

Figure 13 is the key diagram of the number of pixels interconversion rate of embodiment 3.

Figure 14 is the key diagram of the number of pixels interconversion rate of embodiment 3.

Figure 15 is the block scheme of the image-processing system of embodiment 4.

Figure 16 is the figure of an example of the interpolation table of embodiment 4.

Figure 17 is the key diagram of the image processing of embodiment 4.

Figure 18 is the pattern match map table of embodiment 4.

Figure 19 is the pattern match map table of embodiment 4.

Figure 20 is the pattern match map table of embodiment 4.

Figure 21 is the process flow diagram of operation of the preferential interpolating circuit of live width of embodiment 4.

Figure 22 is the image example of the image interpolation of explanation embodiment 4.

Figure 23 is the block scheme of the image-processing system of conventional case.

Figure 24 is the number of pixels conversion figure of expression according to nearest interpolation method and Cubic interpolation method.

The most preferred embodiment of invention

(embodiment 1)

Fig. 1 is the block scheme of the image-processing system of embodiment 1.

These image-processing system 100 inputs are by the picture intelligence (pixel data) of computer 10 outputs, the number of pixels interconversion rate X that sets with the operator carries out after the number of pixels conversion to array display device 20 outputs of PDP, LCD etc., is particularly useful for comprising the occasion of the picture intelligence input of natural image and character and picture.

As Fig. 1, image-processing system 100 is kept interpolating circuit 101, the preferential interpolating circuit 102 of live width, feature line segment detecting circuit 103, is selected circuit 104 and coordinate judgement part 120 etc. to constitute by center of gravity.

Computer 10 sends the pixel data brightness value of matrix coordinate correspondence (image) successively to image-processing system 100, in the explanation, the pixel data of 8 bits of the brightness value (0～255) of each pixel coordinates of expression is sent line by line successively.That is, with coordinate (1,1), (2,1), (3,1) of first row ... order, then with coordinate (1,2), (2,2), (3,2) of second row ... order, from computer 10 input pixel datas.

Coordinate is judged in the part 120, and pixel data of every input is just obtained the suitable pixel coordinates (PQ) (P is a horizontal coordinate, and Q is a vertical coordinate) of which coordinate on this pixel data of expression and the image, sends it to feature line segment detecting circuit 103 etc.

This coordinate is judged the coordinate judgement that part 120 is carried out, for example, can obtain horizontal coordinate P by each pixel data of the signal input back input of the beginning of representing row is counted, the signal of the signal input back by expression one page picture intelligence is begun, the beginning of expression row is counted and is obtained vertical coordinate.

For convenience, at first the situation when only carrying out the number of pixels conversion in the horizontal direction describes.

In the image-processing system 100, be provided with, send number of pixels interconversion rate X to each several part from receiving unit from the receiving unit (not shown) of operator's receiving conversion rate indication.

Feature line segment detecting circuit 103, each picture intelligence to input, according to being the pixel data of the certain limit at center with the target pixel, judge whether this target pixel is the characteristic with feature of character and picture, and this target pixel is " a line segment formation pixel " or " pixel (non-line segment formation pixel) on every side ".In addition, also measure line segment and constitute the continuous in the horizontal direction number W (W is equivalent to line length in the horizontal direction) of pixel.

Center of gravity keeps interpolating circuit 101 and the preferential interpolating circuit 102 of live width all with interconversion rate X the pixel data of common input to be carried out the number of pixels conversion process, and wherein, center of gravity keeps interpolating circuit 101 to adopt making brightness value (" gray-scale value " too.) the number of pixels transform method that after conversion, also remains unchanged of centre of gravity place, specifically, suitable with linear interpolation and Cubic interpolation method etc.

The preferential interpolating circuit 102 of live width is identical with nearest interpolation method basically, after the conversion also former state adopt the brightness value of the pixel of the nearest coordinate position of pixel coordinates before conversion to carry out interpolation, be with the difference of nearest interpolation method, adjust the input pixel data in the line segment formation pixel parts, the number that makes the line segment of output constitute pixel is the number after continuous number W that the line segment to input constitutes pixel multiply by number behind the number of pixels interconversion rate X and rounds, and whole image becomes the conversion image with suitable number of pixels based on number of pixels interconversion rate X.

Here, to " the continuous number W that the line segment of input constitutes pixel multiply by the number behind the number of pixels interconversion rate X " when rounding, can cast out below the radix point also can round-up, round up below the radix point in the present embodiment, (in this instructions, mark [] expression is to the following round of radix point of the number in [] to adopt [continuous number W * interconversion rate X that the line segment of input constitutes pixel].)。

In addition, in the present embodiment, (P, Q) (p q) is equivalent to satisfy the condition (in the formula, if the interconversion rate of vertical direction is 1 to be q=Q, then can only satisfy the condition of formula 1) of following

formula

1,2 to the output coordinate of nearest position from input coordinate.

[(P-1) * X]＜p≤[P * X] ... formula 1

[(Q-1) * X]＜q≤[Q * X] ... formula 2

The picture intelligence that center of gravity keeps interpolating circuit 101 and the preferential interpolating circuit 102 of live width to export respectively carries out the pixel data output in turn line by line after the number of pixels conversion process jointly.In addition, adjust output regularly, make that the picture intelligence after this number of pixels conversion is exported with (interconversion rate) X speed doubly of picture intelligence input speed.

Selection circuit 104 is selected one of them of centers of gravity maintenance interpolating circuits 101 and the preferential interpolating circuit 102 parallel pixel datas of exporting in turn of live width.Promptly, detect the output pixel data for feature line segment detecting circuit 103 into the pixel data correspondence of " characteristic ", selection is from the output of the preferential interpolating circuit 102 of live width, detect the output pixel data for feature line segment detecting circuit 103, select to keep interpolating circuit 101 to handle with center of gravity into the pixel data correspondence of " non-characteristic ".

In addition, though it is not shown, owing to can detect from feature line segment detecting circuit 103, be provided with temporary transient storage in the feature line segment detecting circuit 103 and count the pixel data line storage of the pixel data of row, to the center of gravity picture intelligences of 4 row that kept interpolating circuit 101 and preferential interpolating circuit 102 transmission lags of live width.In addition, to the additional signal that the beginning of 1 page of picture intelligence of expression is arranged and represent the beginning of every row of the picture intelligence of selecting circuit 104 outputs.

(operation of feature line segment detecting circuit 103)

At first, the characteristic that carries out of characterization line segment detecting circuit 103 is judged and an example of line segment decision.

Feature line segment detecting circuit 103 is with the coordinate (P of input pixel, Q) pixel (P of input before 3 row, Q-3) as target pixel, as described below, according to being the pixel data of the certain limit at center with the target pixel, judge whether this target pixel is characteristic (part with feature of character and picture), in addition, judgement is a pixel around line segment formation pixel or the line segment.

Fig. 2 is that characteristic that feature line segment detecting circuit 103 carries out detects and line segment detects the process flow diagram of operation.Pixel data of every input, feature line segment detecting circuit 103 just carries out the processing shown in this process flow diagram.

At first, with reference to the pixel data line storage, (P Q-3) is the interior mean flow rate MA of 3 * 3 pixel coverages at center with target pixel in calculating.

In addition, it is that above group of pixels of mean flow rate MA and brightness are the following group of pixels of mean flow rate MA that the group of pixels in 3 * 3 pixel coverages is divided into brightness, and the group of pixels that order comprises target pixel is group A, and other groups are group B (step S1).

The average brightness MA of calculating group A and coefficient of alteration DA.In addition, the average brightness MB of calculating group B and coefficient of alteration DB (step S2).

Whether investigation organizes the coefficient of alteration DB of the coefficient of alteration DA of A and group B all below 3%, and the difference of average brightness MB of the average brightness MA of group A and group B be more than 64 (step S3, S4).

When satisfying two conditions of above-mentioned steps S3, S4, the judgement target pixel is characteristic (S5).

On the other hand, when the condition of S3, S4 does not satisfy, judge that target pixel is non-characteristic (S6).

When target pixel was judged to be characteristic, as described below, target pixel also carried out the judgement of pixel around the line segment whether.

With the target pixel is the center, judging in than above-mentioned steps S1 in the scope (5 * 5 pixel) of wide ranges of usefulness, is that the number β of the pixel in the number α of the pixel in the scope of MA ± 3% and the scope that brightness is MB ± 3% counts (step S7) to brightness.

If number of pixels α is less than number of pixels β, judge that then target pixel is that line segment constitutes pixel (S8 → S9) is (S8～S10) of pixel around the line segment if number of pixels α, then judges target pixel more than or equal to number of pixels β.

As above-mentioned step S7～S10 α and β are compared, if then decidable α is for constituting the group of pixels of literal for α＜β, this is because usually, when cutting out definite part, the sum of the pixel of the formation background often sum than the pixel that constitutes literal is many.Constitute especially letter etc. of pixel if adopt above-mentioned method to take a decision as to whether line segment, can detect line segment with quite high probability.

In addition, get 5 * 5 pixel coverage judgement in the above-mentioned explanation, empirical evidence, if adopting the above frame of counting of the literal that picture comprised judges, then can detect more accurately, and detect, can improve the precision that detects by getting wideer scope (for example 13 * 13 pixels).

In addition, the method of whether judging with 103 pairs of characteristics of feature line segment detecting circuit is not limited to said method, for example also can adopt being that the spatial frequency in certain zone at center is measured with the target pixel, the spatial frequency of mensuration is judged to be the method for characteristic etc. when high.

In addition, as judging that with feature line segment detecting circuit 103 line segment constitutes the method for pixel, also can consider following method, for example, as described in embodiment 3, detect in the horizontal direction in abutting connection with the big marginal portion of difference of the brightness value of pixel, the occasion that the brightness value of the pixel when between the edge is little judges that the pixel between this edge is that line segment constitutes pixel.

In addition, for example, carry out occasions such as number of pixels conversion, also can consider following method, promptly be provided with and cooperate, can directly know whether be the circuit of line segment formation pixel, whether detected by it is the method that line segment constitutes pixel with the OS of computer at the graphdriver of computer.

Feature line segment detecting circuit 103 is provided with temporary transient storage, and this detects the result (promptly, represent each pixel data whether characteristic characteristic information, represent each pixel data whether line segment constitute the line segment information of pixel) detect result's line storage, capable characteristic information and the line segment information stores till preceding of the pixel data to 4 of input detects the result at this and uses in the line storage now.

Then, the The above results that is detected by feature line segment detecting circuit 103 sends to preferential interpolating circuit 102 of live width and selection circuit 104.

Specifically, owing to import (P now by the pixel data that the preferential interpolating circuit 102 of live width is handled, Q) data of input before pixel data 4 row, thereby feature line segment detecting circuit 103 sends (P, Q-4) line segment information with line storage to the preferential interpolating circuit 102 of live width from detecting the result.At this moment, the number of the characteristic information of output sends after being transformed into (number of pixels interconversion rate) X times.Thereby, select in the circuit 104, can differentiate the pixel data after the number of pixels conversion one by one.

Here, the concrete example according to input image signal describes above-mentioned decision.

Fig. 3 represents the example of input image signal in 5 * 5 pixel coverages that with the target pixel are the center, and (3,3) among this figure are target pixel (coordinates (a, b) be illustrated in horizontal level is that a, upright position are b in 5 * 5 pixel coverages) here.

During Fig. 3 (a) reaches (b), the brightness value of the pattern table aspect element of each pixel coordinates, black picture element is represented near low-light level (brightness value 0), near low-light level during the diagonal line hatches pixel is represented (brightness value 64), the stain shadow pixels is represented near intermediate luminance (brightness value 128), and white pixel is represented near high brightness (brightness value 255).

For example, with the target pixel occasion that forms image shown in Fig. 3 (a) in 5 * 5 pixel coverages at center, owing to be that coefficient of alteration DA or DB get big value in 3 * 3 the group of pixels at center with each pixel, thereby be judged to be non-characteristic.

On the other hand, be in 5 * 5 pixel coverages at center during with the target pixel for visual shown in Fig. 3 (b), if make (2,1), (3,1), (4,1), (2,2), (2,3), (3,3), (4,3), (2,4), (2,5) 9 pixels are group of pixels A, and other 16 pixels are group of pixels B, because the coefficient of alteration DB of the coefficient of alteration DA of group of pixels A and group of pixels B is near 0, line segment constitutes the average brightness MA of group of pixels and the differential nearly 255 of the average brightness MB of group of pixels on every side, thereby the target pixel that can judge this occasion is characteristic, and group of pixels A is judged as line segment and constitutes group of pixels, and group of pixels B is judged as group of pixels around the line segment.

Then, the live width calculating operation that feature line segment detecting circuit 103 is carried out describes.

Fig. 4 is the process flow diagram that feature line segment detecting circuit 103 live width of carrying out is calculated.Feature line segment detecting circuit 103 is carried out the processing shown in this process flow diagram to the pixel data of each input.

(P, whether pixel data Q-4) is that line segment formation pixel begins place (S11) to the coordinate of importing now in the preferential interpolating circuit 102 of feature line segment detecting circuit 103 investigation live widths.If previous coordinate (P-1, line segment information Q-4) is " pixel on every side ", coordinate (P, line segment information Q-4) is " line segment formation pixel ", then these data are the places that begin that line segment constitutes pixel on the horizontal direction.

Then, constitute at line segment under the situation that begins to locate of pixel, obtain thereafter in the horizontal direction line segment and constitute the continuous number (the continuous number W of line segment pixel) of pixel (S11 → S12).

The detection of above-mentioned steps S11 be adopt the investigation of pattern match method detect the result with the pixel coordinates of line storage storage (P, Q-4) and the line segment information of the coordinate of front and back carry out.

The continuous number W of the line segment pixel of obtaining sends (S13) to the preferential interpolating circuit 102 of live width.

And, feature line segment detecting circuit 103 is with reference to the pixel data line storage, coordinate (P, Q-4) continuous afterwards coordinate (P+1, Q-4) pixel data also sends to the preferential interpolating circuit 102 of live width (in addition, this signal is used for the step S28 of the preferential interpolating circuit 102 of live width described later).

(operation of the preferential interpolating circuit 102 of live width)

The preferential interpolating circuit 102 of live width possesses counter T, when the input line segment constitutes pixel, in order to the number of pixel data output after the beginning that constitutes pixel from line segment is counted.

Fig. 5 is the process flow diagram of an example of the operation of expression live width preferential interpolating circuit 102.

The preferential interpolating circuit 102 of live width all carries out as shown in Figure 5 processing to the pixel data of each input.

From the line segment information of feature line segment detecting circuit 103 inputs is the occasion of " pixel on every side ", (being occasion not among the S21) is owing to import (the P of the preferential interpolating circuit 102 of live width now, Q-4) pixel data is " pixel on every side ", thereby carries out nearest interpolation in this occasion and handle (step S21 → S22).

On the other hand, the line segment information of feature line segment detecting circuit 103 inputs is " line segment formation pixel ", and import the occasion of the signal of " the continuous number W of line segment pixel " from feature line segment detecting circuit 103, because now the pixel data of the preferential interpolating circuit 102 of input live width is that line segment constitutes the place that begins of pixel in the horizontal direction, thereby, based on the continuous number W of this line segment pixel, can calculate the line segment formation pixel of representing after the conversion and should export several line segment interpolation number w continuously.This line segment pixel output number w is to the number after (the continuous number W of line segment pixel) * (interconversion rate X) rounds, can be with formula w=[W * X] obtain (S23 → S24).Then, count initialized device T (S25), on counter T, add with input now (the output number of pixels C (satisfying the number of the p of above-mentioned formula 1) that P, pixel data Q-4) have a nearest position relation (S26).

On the other hand, the line segment information that feature line segment detecting circuit 103 sends is " line segment formation pixel ", and do not have an occasion of signal of input " the continuous number W of line segment pixel " from feature line segment detecting circuit 103, mean that then this pixel is that line segment constitutes the follow-up pixel of pixel in advance, thereby count initialized device T not, on counter T, add with input now (P, pixel data Q-4) have the output number of pixels C (S23 → S26) of nearest position relation.

Then, by carrying out the processing of following steps S27～S30, output constitutes pixel with the line segment of [the continuous number W of line segment pixel * interconversion rate X] suitable number continuously.The value of counter T is if surpass the value of w, when C conversion pixel data of output, individual with (C-T+w) that the output of nearest interpolation method is initial, remaining (T-w) individual then to be adopted and next coordinate (the P+1, (data of sending from feature line segment detecting circuit 103 of pixel data Q-4).) the brightness value output (S27 → S28) that coincide.

On the other hand, the value of counter T is if be less than or equal to the value of w, input now (P, pixel data Q-4) they are not W occasions (occasion of the pixel between W continuous pixel), carry out nearest interpolation (S27 → S29 → S22).

In addition, the value of counter T is if be less than or equal to the value of w, input now (P, pixel data Q-4) they are W occasions (occasion of the pixel that W continuous pixel is last), carry out nearest interpolation, also export (w-T) individual pixel data with same brightness simultaneously.In addition, in this occasion, for follow-up input (when the nearest interpolation of step S22, it is individual to skip (w-T) that start in the output coordinate of proximal most position for P+1, pixel data Q-4), begins to carry out nearest interpolation (S27 → S28 → S30) from its follow-up coordinate.

The preferential interpolating circuit 102 of live width, by carrying out the processing of above-mentioned steps S27～S30, constitute the occasion of pixel or import occasion more than 2 continuously only importing line segment, all continuously output and [the continuous number W of line segment pixel * interconversion rate X] quite the line segment of number constitute pixel.

(selecting the operation of circuit 104)

In selecting circuit 104, when keeping pixel data after interpolating circuit 101 and the preferential interpolating circuit 102 parallel input conversion of live width from center of gravity, send result of determination (target pixel whether characteristic) from feature line segment detecting circuit 108 synchronously, thereby, as described below, according to the result of determination of feature line segment detecting circuit 108, select a wherein side pixel data of output.

That is, select circuit 104 to all processing shown in the process flow diagram of execution graph 6 of the pixel data of each input.

Then, when the result of determination of feature line segment detecting circuit 103 is " non-characteristic ", the pixel data of preferential interpolating circuit 102 outputs of selection live width (step S31 → S32), during for " characteristic ", the pixel data of selection center of gravity maintenance interpolating circuit 101 outputs (step S31 → S33).

(also carrying out the occasion of number of pixels conversion in vertical direction)

More than the situation of only carrying out the number of pixels conversion in the horizontal direction is illustrated, as above-mentioned, after carrying out the conversion of horizontal direction with horizontal direction interconversion rate XH, also can carry out conversion with number of pixels interconversion rate XV equally in vertical direction, thus can be in the enterprising line translation of horizontal vertical both direction.

For example, as above-mentioned, with horizontal direction interconversion rate XB in the horizontal direction behind the up conversion, to this output image signal, the order of pixel data is become row from line translation, use above-mentioned same device then, XV carries out conversion with the vertical direction interconversion rate.

Then, to this output image signal, by once again with the order of pixel data from the line translation to row, can carry out the number of pixels conversion with horizontal direction interconversion rate XH, vertical direction interconversion rate XV.

Or, utilize the result that detects of feature line segment detecting circuit 103, carry out the conversion of vertical direction when carrying out the conversion of above-mentioned horizontal direction, can carry out conversion at the horizontal vertical both direction.

In this occasion, when feature line segment detecting circuit 103 is also measured the continuous number W of line segment pixel of vertical direction and when selecting circuit 104 to send characteristic informations, the number of the characteristic information of output is transformed into (number of pixels interconversion rate XH * number of pixels interconversion rate XV) doubly, and feature line segment detecting circuit 103 is carried out the same processing of process flow diagram with above-mentioned Fig. 4 in vertical direction.In addition, the preferential interpolating circuit 102 of live width is carried out in vertical direction and the same processing of process flow diagram of above-mentioned Fig. 5.But be necessary to change, a plurality of pixels that are 1 row are provided with the counter Tv of vertical direction, and interlacing is counted in each counter Tv.

(effect of present embodiment)

At first, to the concrete example of input imagery, relatively center of gravity keeps interpolation method, the preferential interpolation method of live width and nearest interpolation method.

Fig. 7 represents that input imagery when vertical direction interconversion rate XV=1 carries out the number of pixels conversion (8 pixels * 4 pixels) is transformed into the situation of output image (4 pixels * 18 pixels) with horizontal direction interconversion rate XH=2.25.Input imagery shown in Fig. 7 (a)～(c) all is that the line segment of counting the 3rd (H3) and the 6th (H6) coordinate from the left side constitutes pixel, has the part of the feature of character and picture.

(a) expression keeps occasion behind this input imagery of interpolation method conversion with center of gravity, and in this occasion, 4 the pixel interpolations of pixel before by conversion of respectively exporting after the conversion form.Though the centre of gravity place of brightness remains unchanged after conversion, the line segment marginal portion thickens after the conversion.

(b) expression is carried out occasion after the conversion with the preferential interpolating circuit 102 of above-mentioned live width, in this occasion, the line segment of H3 in the input coordinate (wide 1 pixel of line segment) is transformed into the line segment (wide 2 pixels of line segment) of D6 in the output coordinate, D7, and the line segment of the H6 in the input coordinate (wide 1 pixel of line segment) also is transformed into the line segment (wide 2 pixels of line segment) of D12 in the output coordinate, D13.

(c) represent with the occasion after the nearest interpolation method conversion, at this moment, the line segment of H3 in the input coordinate (wide 1 pixel of line segment) is transformed into the line segment (wide 2 pixels of line segment) of D6 in the output coordinate, D7, and the line segment of the H6 in the input coordinate (wide 1 pixel of line segment) is transformed into the line segment (wide 3 pixels of line segment) of D12 in the output coordinate, D13, D14.

Thereby, part to feature with character and picture, if adopt the preferential interpolating circuit 102 of live width to carry out interpolation, then the marginal portion of line segment can not thicken, and the line segment that has same live width in the input imagery on the horizontal direction also keeps same after conversion, and marginal position also before conversion near the position.

In addition, when carrying out conversion in vertical direction, can keep the homogeneous of the live width after the number of pixels conversion in the vertical direction too.Thereby for the either direction in the horizontal direction vertical direction oblique line directions, the line segment of same live width can both keep same in the input imagery after the number of pixels conversion.

In addition, because curve can be regarded as the line that is formed by connecting by short line segment, thereby the curve of same live width also can keep same live width in the input imagery after the number of pixels conversion.

In addition,, compare, can be offset because of interconversion rate causes the position of line segment, thereby may produce not harmony sense with nearest interpolation method if carry out conversion with the preferential interpolation method of live width.But, in character and picture, in general, whether there is skew to compare with the position of line segment, whether live width keeps the influence of all a pair of not harmony sense bigger.

Particularly, if the line segment width that comprises in the image of ruling etc. is inconsistent, the very not harmony that then can seem, and by the preferential interpolation method of live width in the present embodiment can keep the homogeneous of live width behind the image transform, thereby can obtain very practical effect.

In addition, in the present embodiment, owing to whether have the characteristic of character and picture and alternately adopt preferential interpolating circuit 102 of live width and center of gravity to keep interpolating circuit 101 according to target pixel, thereby, the part suitable with character and picture (literal and ruling are main image) adopted the preferential interpolating circuit 102 of live width, then the line segment part can not thicken, and it is same that the line segment of the same live width in the input imagery keeps after the number of pixels conversion, on the other hand, select centre of gravity place to keep interpolating method to the part suitable, can obtain to have the output image of natural image harmony sense with natural image.

(embodiment 2)

Fig. 8 is the block scheme of the image-processing system of embodiment 2.Same with the image-processing system 100 of embodiment 1, image-processing system 200 usefulness number of pixels interconversion rate X (the horizontal direction interconversion rate XH of present embodiment, vertical direction interconversion rate XV) picture intelligence from computer 10 outputs is carried out the number of pixels conversion and outputs to display device 20, its difference is, carry out conversion at number of pixels interconversion rate X near the preferential interpolation of occasion employing live width of integer, the occasion that number of pixels interconversion rate X keeps off integer adopts center of gravity to keep interpolation to carry out conversion.

For convenience of explanation, the situation of only carrying out the number of pixels conversion in the horizontal direction is described at first in the present embodiment.

This image-processing system 200 is kept interpolating circuit 201, the preferential interpolating circuit 202 of live width, line segment detecting circuit 203, is selected circuit 204 and coordinate judgement part 220 etc. to constitute by center of gravity.

Wherein, center of gravity keeps interpolating circuit 201, the preferential interpolating circuit 202 of live width and coordinate to judge that part 220 keeps interpolating circuit 101, the preferential interpolating circuit 102 of live width and coordinate to judge that part 120 is identical with the center of gravity of embodiment 1 explanation.

The line segment detecting circuit 203 also feature line segment detecting circuit 103 with embodiment 1 is identical.But, in the present embodiment, because selecting circuit 204 selects according to number of pixels interconversion rate X, line segment detecting circuit 203 is to selecting circuit 204 to send the characteristic information of expression " target pixel whether characteristic ", but by receiving unit to selecting circuit 204 to send the signal of number of pixels interconversion rate X.

(selecting the operation of circuit 204)

Fig. 9 is the process flow diagram that the operation of circuit 204 is selected in expression, to the pixel data execution processing as shown in the figure of each input.

Select circuit 204 if receive result of determination from feature line segment detecting circuit 103 (target pixel whether characteristic), then when the pixel data that at every turn keeps after the preferential interpolating circuit 102 parallel input conversion of interpolating circuit 101 and live width from center of gravity, according to result of determination, select one of them also output from 103 inputs of feature line segment detecting circuit.

This selection operation is shown in the process flow diagram of Fig. 9, select pixel data (step S41 → S42) during the value " near integer " of the number of pixels interconversion rate X that receiving unit sends from preferential interpolating circuit 102 outputs of live width, when " keeping off integer ", and the pixel data that selection is exported from center of gravity maintenance interpolating circuit 101 (step S41 → S43).

As judging whether number of pixels interconversion rate X can enumerate following example near integer-valued method, specifically, the occasion of the difference of interconversion rate X and [X] below the value Δ of regulation (| X-[X] | the occasion of≤Δ) judge " number of pixels interconversion rate X is near round values " (the value Δ of this regulation be less than 0.5 0.3,0.2 or 0.1.), during greater than the value Δ of regulation (below | X-[X3]≤occasion of Δ) judge " number of pixels interconversion rate X keeps off round values ".

In addition, the situation of only carrying out the number of pixels conversion in the horizontal direction has been described more than, identical with embodiment 1, if the conversion of level of aggregation direction and the conversion of vertical direction then can be carried out conversion at the horizontal vertical both direction.

(effect of present embodiment)

According to the number of pixels conversion of present embodiment, during near round values, select the number of pixels transform method at number of pixels interconversion rate X, make to keep the sets of line segments of same live width after conversion, also can keep consistent with each other in the input imagery.

In this occasion, for the character and picture part, the line segment part can not thicken, and the line segment of same live width also keeps same live width in the input imagery after the number of pixels conversion.In addition, for natural image, though select the preferential interpolating method of live width, as number of pixels interconversion rate X during near round values, even, also can not cause serious not harmony sense to the preferential interpolating method of natural image employing live width.

On the other hand, when number of pixels interconversion rate X keeps off integer, select the centre of gravity place of Cubic interpolation method and linear interpolation etc. to keep interpolating method.

In this occasion, natural image is partly adopted suitable conversion, the character and picture part is also become adopt center of gravity to keep interpolating method to carry out conversion.

In the preferential interpolating method of live width, when number of pixels interconversion rate X keeps off integer, because the offset of line is bigger, thereby be easy to produce not harmony sense, in the present embodiment, when keeping off integer, number of pixels interconversion rate X,, can suppress the not harmony sense of the offset generation of line though then lines thicken if adopt centre of gravity place to keep interpolating method to carry out conversion.

(variation of present embodiment)

Can cumulated volume embodiment 2 and embodiment 1.

For example, can adopt following system of selection, promptly, as above-mentioned embodiment 1, detect whether characteristic of pixel data with line segment detecting circuit 203, number of pixels interconversion rate X is during near integer, the unconditional output of selecting the preferential interpolating circuit 202 of live width, on the other hand, when number of pixels interconversion rate X keeps off integer, the same with embodiment 1, if pixel data is a characteristic, then, when keeping off integer, number of pixels interconversion rate X select center of gravity to keep the output of interpolating circuit 201 with selecting circuit 204 to select the output of the preferential interpolating circuit 202 of live width.

(embodiment 3)

Figure 10 is the block scheme of the image-processing system of embodiment 3.The same with the image-processing system 100 of embodiment 1, the image-processing system 300 of present embodiment also uses number of pixels interconversion rate X (horizontal direction interconversion rate XH, vertical direction interconversion rate XV) that the picture intelligence from computer 10 outputs is carried out the number of pixels conversion and outputs to display device 20.But, in the present embodiment, to being equivalent to the pixel data of marginal portion, to carry out the number of pixels conversion from the nearest integral multiple of number of pixels interconversion rate X, on the other hand, the pixel data beyond the marginal portion is adjusted, made in the not serious scope that departs from number of pixels interconversion rate X, and make whole output image carry out the number of pixels conversion, so that carry out the preferential interpolation of live width with number of pixels interconversion rate X.

This image-processing system 300 judges that by marginal portion detecting circuit 303, the preferential interpolating circuit 310 of live width and coordinate part 320 constitutes, and wherein, coordinate judges that part 320 is identical with the coordinate judgement part 120 of embodiment 1.

Marginal portion detecting circuit 303 judges according to the luminance difference of the pixel of the brightness of target pixel and adjacency whether this target pixel is equivalent to the edge and constitutes pixel, and this result of determination sends to the preferential interpolating circuit 310 of live width as marginal information.

The preferential interpolating circuit 310 of live width carries out interpolation according to nearest interpolation method basically, but the marginal information of sending according to marginal portion detecting circuit 303, at the pixel data of input is that the occasion that the edge constitutes pixel is adjusted, and the number that makes the edge of output constitute pixel is integer [number of pixels interconversion rate X].In addition, be that non-edge constitutes pixel place adjustment output number of pixels at the pixel data of importing, make whole output image signal export with number of pixels interconversion rate X.

As shown in figure 10, in the preferential interpolating circuit 310 of such live width, integral multiple enlarged 311 is set, any several times enlarged 312 and selection part 313, the preferential interpolating circuit 310 of live width can be by selecting part 313 to carry out switching controls, when being edge formation pixel, adopts the pixel data of input integral multiple enlarged 311, generate [number of pixels interconversion rate X] individual output pixel with the brightness value identical with the pixel data of input, when being non-edge formation pixel, adopts the pixel data of input any several times enlarged 312, adjust output number of pixels and output, in addition, according to the concrete example of the operation of marginal portion detecting circuit 303 shown below and the preferential interpolating circuit 310 of live width, also can realize same function.

(operation of marginal portion detecting circuit 303)

Although it is not shown, but in order in marginal portion detecting circuit 303, to detect, be provided with the line storage of counting the pixel data of row in order to temporary transient storage in the marginal portion detecting circuit 303, to preferential interpolating circuit 310 transmission lags of live width 2 the row picture intelligences.

Marginal portion detecting circuit 303 is judged whether marginal portion of target pixels, measures the edge simultaneously and constitutes the continuous in the horizontal direction number Z of pixel.

The marginal portion decision of marginal portion detecting circuit 303 at first, is described.

Marginal portion detecting circuit 303 is with the coordinate (P of the pixel of input, Q) (P is Q-1) as target pixel, as described below for the pixel of input before 1 row, by the pixel data that the front and back of comparison object pixel are imported, judge whether the edge constitutes pixel to this target pixel.

Specifically, can adopt following method, promptly, the difference of the brightness value of the pixel of the front and back adjacency of target pixel is if when (for example more than 64, or more than 32), be judged to be the marginal portion more than the value of regulation, the difference of this brightness value then is judged to be non-marginal portion as if the value less than afore mentioned rules.

In addition, also can adopt following method, that is, judge with the target pixel be the spatial frequency of near zone at center whether more than the value of regulation, when the difference of brightness value and these two conditions of spatial frequency all satisfy, be judged to be the marginal portion.

Judge by this, can in the horizontal direction the pixel that forms the marginal portion of figure, literal, ruling etc. be constituted pixel as the edge and detect.

In addition, marginal portion detecting circuit 303 is provided with temporary transient storage, and this detects the result (promptly, represent that whether each pixel data constitutes the marginal information of pixel in the edge) detect result's line storage, will be from the pixel data to 2 of present input capable preceding marginal information be stored in this and detect result's line storage.

The continuous edge number calculating operation of marginal portion detecting circuit 303 then, is described.

Figure 11 is the process flow diagram that the continuous edge number of marginal portion detecting circuit 303 calculates.Marginal portion detecting circuit 303 is carried out the processing shown in this process flow diagram to the pixel data of each input.

This live width calculating operation of handling the feature line segment detecting circuit 103 that illustrates with embodiment 1 is roughly the same, below carries out simple declaration.

Now, (whether P, pixel data Q-2) edge pixel (edge constitutes pixel end) (S51) for the coordinate of the preferential interpolating circuit 310 of investigation input live width.

(P, marginal information Q-2) is that (P+1 when line segment information Q-2) is " non-edge formation pixel ", judges in the horizontal direction that then it is end that this edge constitutes pixel for " edge formation pixel ", next coordinate to coordinate.

Then, it is the occasion of end that the edge constitutes pixel, obtains the continuous number (the continuous number Z of edge pixel) of the edge formation pixel that arrive in the horizontal direction till it (S51 → S52).

Above-mentioned steps S51 detects, for example, can by detect with pattern match method investigation the result with the pixel coordinates of line storage storage (P, Q-2) and the marginal information of the coordinate of front and back carry out.

Then, the continuous number Z of the edge pixel of obtaining is sent (S53) to the preferential interpolating circuit 310 of live width.

And feature line segment detecting circuit 103 is with reference to the pixel data line storage, and (P, (P+1, pixel data Q-2) also send to the preferential interpolating circuit 310 of live width subsequent coordinates Q-2) with coordinate.

(operation of the preferential interpolating circuit 310 of live width)

Be provided with counter T in the preferential interpolating circuit 102 of live width, when constituting pixel, in order to the number that constitutes the pixel data of exporting after pixel begins from the edge is counted at the input edge.

Figure 12 is the process flow diagram of an example of the operation of expression live width preferential interpolating circuit 310, and the preferential interpolating circuit 310 of live width is carried out this processing shown in Figure 12 to the pixel data of each input.This operation is as follows.

The marginal information of marginal portion detecting circuit 303 input be the occasion of " non-edge formation pixel ", carries out nearest interpolation processing (step S61 → S62) basically.

On the other hand, if the marginal information " edge formation pixel " of feature line segment detecting circuit 103 inputs, then the conversion pixel data is exported (M is the value after interconversion rate X is rounded) so that every integer M is individual.Here, make M=[interconversion rate X] and (in the formula, some M=1 of 0＜X＜0.5) (S61 → S63).

In counter T, add with input now (the output number of pixels C (satisfying the number of the p of above-mentioned 1 formula) that P, pixel data Q-2) have a nearest position relation (S65).

Then, (be not occasion not among the step S65) when the signal of " the continuous number Z of edge pixel " is imported, showing has follow-up edge to constitute pixel, thereby operation directly finishes.

On the other hand, when the signal input of " the continuous number Z of edge pixel " is arranged, show that then edge in the horizontal direction constitutes the pixel end, adjusts to following step S66～S68.

The value of counter T if more than or equal to (M * Z), then only output (T-MZ) individual with subsequent coordinates (P+1, the pixel data of the brightness value correspondence of pixel data Q-2) (S66 → S67).

On the other hand, if the value of counter T is less than (M * Z), (P+1 for follow-up input, Q-2) pixel data, when in step S62, carrying out nearest interpolation, skip in the output coordinate of proximal most position (MZ-T) of beginning individual after, begin to carry out nearest interpolation (S66 → S68) from thereafter coordinate.

The preferential interpolating circuit 310 of live width constitutes pixel by carrying out above-mentioned processing for the edge, and output and [interconversion rate X] suitable number constitute pixel pixel in addition for the edge, adjust the output number of pixels, make whole output image carry out conversion with interconversion rate X.

But, in the input imagery, constituting pixel and pixel is with the fine pitches place arranged side by side of 1 pixel or 2 pixels on every side at line segment, the continuous number Z of edge pixel also may become quite big.

Thereby, the big occasion of the continuous number Z of edge pixel, and the value of interconversion rate X keeps off the occasion of integer, if carry out above-mentioned processing, then marginal position in the input imagery and the marginal position in the output image significantly are offset, and very not harmony seems.

Thereby, for fear of above-mentioned shortcoming, in the feature line segment detecting circuit 103, the maximal value that makes the continuous number Z of edge pixel is 4, when 5 or more edge formation pixel consecutive hours, also can constitute pixel and handle as non-edge formation pixel exceptionally the 5th.

(conversion on the vertical direction)

More than in the horizontal direction the situation of number of pixels conversion only is illustrated, but, the same with the explanation of embodiment 1, if the conversion of level of aggregation direction and the conversion of vertical direction, though must change to design, for example use the continuous number Z of edge pixel of marginal portion detecting circuit 303 mensuration vertical direction etc., but can be at horizontal vertical both direction up conversion.

Below, the number of pixels interconversion rate of present embodiment is described with regard to the concrete example of input imagery.

Figure 13 represents with horizontal direction interconversion rate XH=2.25, when vertical direction interconversion rate XV=1 carries out the number of pixels conversion, in the horizontal direction the input image signal of input coordinate (H1 is to H8) is carried out the number of pixels conversion according to said method, and output to the output coordinate (situation of D1～D18).

Among the figure, and the input coordinate that white shows (H1, H2, H6, H7, H8) brightness value of all representing pixel data is near 255, and (H3, H4 H5) represent that all the brightness value of pixel data is near 0 to the input coordinate of black.

This occasion, the pixel data of 4 input coordinate H2, H3, H5, H6 constitute pixel as the edge and are detected.

The occasion of interconversion rate X=2.25 is because M=[2.25]=2, as shown in figure 13, and input coordinate (H2, H3, H5, input pixel data H6) outputs to closer position (function of integral multiple enlarged 311) with 2 times number of pixels.

On the other hand, the pixel data of H1, H4, H7, H8 carry out conversion and output with nearest interpolation method basically, and whole output number of pixels is adjusted into 18, makes all interconversion rates become 2.25.

(effect of present embodiment)

As previously discussed, according to the number of pixels interconversion rate of present embodiment, the marginal portion of input imagery can not fog after conversion yet.In addition, image and interconversion rate X according to input though the position of the marginal portion after the conversion may be offset, do not have significant skew usually.

In addition, live width is the image area of 1 pixel or 2 pixels, uses the certain amplification degree M that is determined by interconversion rate X to carry out conversion.

That is, in the input imagery, the continuous number that line segment constitutes pixel is the occasion of 1 (live width is the image area of 1 pixel), and the continuous number Z of edge pixel is generally 3, this occasion, and the line segment after the conversion constitutes pixel and exports continuously with M.In addition, the continuous number of line segment formation pixel is 2 occasion, and the continuous number Z of edge pixel is generally 4, this occasion, and the line segment after the conversion constitutes pixel and exports continuously with 2M.

Thereby to the image that comprises fine rule literal and fine rule figure that computer etc. often uses, it is inconsistent that the live width of fine rule can not become, and conversion image than harmony can obtain to seem.

In addition, in the number of pixels interconversion rate of present embodiment, line segment constitutes the occasion of continuous number more than 3 of pixel, and line segment constitutes in the beginning part of pixel, the continuous number Z of edge pixel is generally 2, in this occasion, the live width after the conversion may not become necessarily, still, for big live width, even live width is more or less inconsistent, the very not harmony that also can not seem, thereby be very practical for the live width of the fine rule of only keeping 1 pixel, 2 pixels.

In addition, be not limited to horizontal direction, in vertical direction, for the image area of thin live width, live width also keeps certain after the number of pixels conversion equally.That is, for the either direction of horizontal direction vertical direction oblique line directions, the fine rule section of same live width also can be consistent after the number of pixels conversion in the input imagery, and the curve of same live width also can be consistent after the number of pixels conversion in the input imagery.

(variation of present embodiment)

Basically adopt the identical method of brightness value of the brightness value and the pixel data after the conversion of input pixel data in the above-mentioned explanation, but shown in the example of Figure 14, can adopt basically with above-mentioned the same integral multiple expansion the marginal portion and carry out interpolation, and also adopt linear interpolation to carry out interpolation leaving the edge part office.

Promptly, in the example of Figure 14, D4～D7, D11～D14 of the position of marginal portion in the output coordinate near is adopted integral multiple to enlarge to carry out interpolation (among Figure 14 shown in the solid arrow), and brightness value in statu quo distributes, for D1～D3, D8～D10, the D16～D18 of the position of leaving the marginal portion in the output coordinate, carry out interpolation according to the linear interpolation of 2 branches (among Figure 14 shown in the dotted arrow)?

Such number of pixels conversion can followingly be carried out.

In the above-mentioned image-processing system 300, be set up in parallel the linear interpolation circuit, be provided with the selection circuit simultaneously, be used to select the output of the preferential interpolating circuit 310 of live width and the output of this linear interpolation circuit with the preferential interpolating circuit 310 of live width.Then, select circuit to send marginal information from marginal portion detecting circuit 303 to this, marginal information is the occasion of " marginal portion ", selects the output of the preferential interpolating circuit 310 of live width, marginal information is the occasion of " non-marginal portion ", selects the output of linear interpolating circuit.

In addition, the example of above-mentioned Figure 14 forms with the linear interpolation of 2 branches, if there is not the restriction of hardware, then the number of branch can be arbitrarily, can certainly use the Cubic interpolation.

Thereby, by and with linear interpolation and Cubic interpolation method, except above-mentioned line segment can not thicken and keep the effect of homogeneity of live width, can also improve the image quality of the image after the conversion.That is, because the increase of the flatness beyond the marginal portion, thereby the sense naturally of image quality improves.In general, adopt the Cubic interpolation than the better effects if that adopts linear interpolation, and branch the multiple-effect fruit is good more more.

(embodiment 4)

Figure 15 is the block scheme of the image-processing system of embodiment 4.

This image-processing system 400 keeps interpolating circuit 401, characteristic detecting circuit 404, B to select circuit 406, the preferential interpolating circuit 410 of live width and coordinate transform part 420 etc. to constitute by center of gravity.

Wherein, center of gravity keeps interpolating circuit 401 to keep interpolating circuit 101, the preferential interpolating circuit 102 of live width and coordinate to judge that part 120 is identical with coordinate transform part 420 with the center of gravity of embodiment 1 explanation.

The characteristic detecting circuit 404 also feature line segment detecting circuit 103 with embodiment 1 explanation is identical, picture intelligence to each input, according to being the pixel data of the certain limit at center with the target pixel, judge whether characteristic of this target pixel, and judge that this target pixel is " a line segment formation pixel " or " pixel (non-line segment formation pixel) on every side ", select circuit 406 to send to preferential interpolating circuit 410 of live width and B this result of determination.

Possess nearest interpolating circuit 411, pattern match interpolating circuit 412 and A in the preferential interpolating circuit 410 of live width and select circuit 413,, carry out interpolation with nearest interpolating circuit 411 and pattern match interpolating circuit 412 and handle according to number of pixels interconversion rate signal X.

Recently interpolating circuit 411 adopts and will carry out interpolation from the method for the brightness value of the pixel after the brightness value former state of the input pixel of the nearest coordinate position of the pixel after the conversion is as conversion.Pattern match interpolating circuit 412 is that each pixel at input image signal is in certain zone at center, and whether according to line segment constitutes the information of pixel, carries out the circuit of interpolation with the gimmick of pattern match.In addition, the back will be described in detail, that is, pattern match interpolating circuit 412 has the interpolation table that each number of pixels interconversion rate X is set, and carries out interpolation with reference to this interpolation table.A selects to possess line storage in the circuit 418, is carried out the picture intelligence after the number of pixels conversion or is carried out pixel data after the number of pixels conversion by pattern match interpolating circuit 412 by nearest interpolating circuit 411 in order to temporary transient storage.Then, the picture intelligence after the number of pixels conversion of the temporary transient storage of this line storage is exported in turn.

Picture intelligence after the conversion of above-mentioned center of gravity maintenance interpolating circuit 401 outputs, and the picture intelligence after the conversion of A selection circuit 413 outputs send the pixel after the conversion simultaneously one by one to B selection circuit 406.In addition, adjust output regularly, make that the picture intelligence after this number of pixels conversion is exported with (horizontal direction interconversion rate XH) * (the vertical direction interconversion rate XV) speed doubly of picture intelligence input speed.

The result of determination (whether being the result of determination of characteristic) that B selection circuit 406 bases are sent by characteristic detecting circuit 404, one of them in the pixel datas of selecting to export side by side by center of gravity maintenance interpolating circuit 401 and the preferential interpolating circuit 410 of live width.

(interpolation table of pattern match interpolating circuit 412)

Figure 16 represents an example of the interpolation table that the preferential interpolating circuit 410 of live width adopts, when the interpolation table of this illustrated example is used for number of pixels interconversion rate X for (the interconversion rate XH=1.2 of horizontal direction times, the interconversion rate XV=1.4 of vertical direction doubly).

Pattern match interpolating circuit 412, based on proximal most position relation or near pixel data before the conversion of position, identical on the pixel data this point after the generation conversion with nearest interpolating circuit, its difference is, shall not be applied to the pixel coordinates after near the conversion of position all as the brightness value of the pixel of interpolation benchmark, and just be applied to the pixel coordinates selected according to pattern match, by adopting this mode to carry out interpolation, can carry out live width is also kept homogeneous and do not had lines to interrupt after conversion number of pixels conversion.

The ratio of the number of pixels of input imagery and the output image corresponding with it is determined by number of pixels interconversion rate X.This number of pixels interconversion rate X is that (the interconversion rate XH=1.2 of horizontal direction doubly, the interconversion rate XV=1.4 of vertical direction is doubly) occasion, the pictorial data of 5 pixel * 5 pixels of input (25 pixel datas) is transformed into the pixel (42 pixel datas) of 6 pixel * 7 pixels.

Figure 17 is the figure of the above-mentioned interpolation table of explanation, and is identical with the interpolation table of Figure 16, the coordinate range of expression input imagery and have the coordinate range of the output image of same position with it in picture.

Among this figure, near the pixel coordinates of position is one by one at once in each pixel coordinates of importing pixel coordinates scope (25 pixels) and output pixel coordinates scope (42 pixels), (in the output pixel coordinates of Figure 17, " residue pixel coordinates " represented with diagonal line hatches then to have produced non-" the residue pixel coordinates " one to one of (42-25)=17 pixel in the output image coordinate.)。

With the input pixel coordinates (R, S) export one to one pixel coordinates (r, s) through type r=[R * XH], s=[S * XV] obtain.

For example, because the horizontal coordinate of exporting pixel coordinates one to one with input pixel coordinates (3,3) is that [3 * 1.2]=4, vertical coordinate also are [3 * 1.4]=4, thereby become (4,4).

Then, each coordinate of this " residue pixel coordinates " is selected input pixel coordinates as the interpolation benchmark, promptly, (1,1), (2,1), (4,1), (5,1), (6,1), (2,3), (1,4), (2,4), (4,4), (5,4), (6,4), (2,6), (2,7) totally 13 pixel coordinates, in the input pixel coverage of Figure 17, use the horizontal line shadow representation.

In the interpolation table of Figure 16, for the pixel data corresponding with the coordinate of this interpolation benchmark, also regulation adopts pattern match.

Promptly, in the interpolation table of Figure 16, with table reference coordinates (R, S) place of suitable coordinate range, be recited as the occasion of " vertically ", " level " or " both direction ", mean in pattern match interpolating circuit 412 by going up expansion in " vertical direction " " horizontal direction " or " both direction ", select the output pixel coordinates of record in the coordinate range to carry out the interpolation processing, in the interpolation table of Figure 16, with table reference coordinates (R, S) suitable coordinate range is blank occasion, means the output of selecting nearest interpolating circuit 411.

In addition, in the output pixel coordinates scope of Figure 17, with the pixel of interpolation the benchmark left side or the top adjacency of coordinate (using the vertical line shadow representation) and " residue pixel coordinates " one to one.

In addition, shown in the output image coordinate range of Figure 17, the 2nd row and the 5th row all are equivalent to " residue pixel coordinates ", they with the input pixel data of the pixel coordinates correspondence of the 1st row in the input pixel coordinates scope and the 3rd row as the interpolation benchmark.

As mentioned above, but above-mentioned interpolation table set interpolation for and export pixel coordinates one to one with the pixel that becomes the interpolation benchmark, simultaneously, also can interpolation be positioned near it " residue coordinate " of (right, down or bottom right).

In addition, the interconversion rate XH=1.2 that horizontal direction has been described here doubly, the interconversion rate XV=1.4 of vertical direction interpolation table doubly, but, can set interpolation table too for interconversion rate arbitrarily.

(operation of characteristic detecting circuit 404)

Characteristic detecting circuit 404, coordinate (P with the input pixel, Q) pixel (P that the suitable time of 3 row is imported before, Q-3) be target pixel, according to being the pixel data of the certain limit at center with the target pixel, whether judge this target pixel in characteristic inside, in addition, judgement is a pixel around line segment formation pixel or the line segment.

Characteristic that this characteristic detecting circuit 404 carries out judges and the line segment decision, and is identical with the operation of the characteristic detecting circuit 108 of the flowchart text of Fig. 2 among the reference embodiment 1, omits its explanation here.

In addition, detect in order to carry out this, similarly to Example 1, be provided with the pixel data line storage of temporary transient storage pixel data in the characteristic detecting circuit 404, in addition, also by characteristic detecting circuit 404 to the center of gravity picture intelligences of 5 row that kept interpolating circuit 401 and preferential interpolating circuit 410 transmission lags of live width.

The result who is detected by characteristic detecting circuit 404 sends to pattern match interpolating circuit 403 and B selection circuit 406.

Specifically, because the pixel data of handling in the pattern match interpolating circuit 403 is with respect to (the P of present input, Q) pixel data is the 5 row data of input before, thereby send with (P to pattern match interpolating circuit 403 with line storage from detecting the result, Q-5) be the line segment information of 3 * 3 pixel coverages at center, as the object pattern-information of pattern match.Meanwhile, line segment constitute the average brightness MA of group of pixels and on every side the average brightness MB of group of pixels also to 403 transmissions of pattern match interpolating circuit.

In addition, read with line storage from detecting the result that (P, characteristic information Q-5) also selects circuit 406 to send to B.

In addition, when B selects circuit 406 to send characteristic informations,, the number of characteristic information is transformed into (horizontal direction interconversion rate XH) * (vertical direction interconversion rate XV) sends again after doubly according to number of pixels interconversion rate X.Promptly, sending to B selects the number of the result of determination information of circuit 406 to multiply by horizontal direction interconversion rate XH, for with the input pixel coordinates of above-mentioned Figure 17 in become the 1st row of interpolation benchmark and the corresponding decision signal of pixel coordinates of the 3rd row, select circuit 406 to send to B 2 times repeatedly.Thereby, in B selects circuit 406, can differentiate the pixel data after the number of pixels conversion one by one.

(operation of the preferential interpolating circuit 410 of live width)

Here, for convenience of explanation, to the number of pixels interconversion rate X by operator's appointment is (horizontal direction interconversion rate XH=1.2, vertical direction interconversion rate XV=1.4), the width that the line segment that detects with characteristic detecting circuit 404 constitutes pixel is that the situation (that is, line segment constitute one of the continuous number of horizontal direction of pixel or the continuous number of vertical direction be 1 occasion) of a pixel describes.

When Figure 18～Figure 20 represents that the live width that is detected by characteristic detecting circuit 404 is 1 pixel, the map table that pattern match interpolating circuit 412 uses.In addition, the back will describe this map table in detail.

The preferential interpolating circuit 410 of live width carries out the processing shown in the process flow diagram of Figure 21 to the pixel data of each input.

The line segment information that characteristic detecting circuit 404 sends is selected the output (S72) of nearest interpolating circuit 411 for the occasion (being occasion not among the S71) of " non-line segment formation pixel ".

On the other hand, the line segment information that characteristic detecting circuit 404 sends is the occasion of " line segment formation pixel ", and (P, Q-5), circulation is with reference to the interpolation table of Figure 16, the decision interpolating method according to the coordinate of target pixel.

Thereby, at first, table reference coordinates (R, S) (S73) of the target pixel that adopts when calculating the interpolation table with reference to Figure 16.

In the interpolation table of Figure 16, because the number of pixels of horizontal direction is 5, vertical number of pixels also is 5, thereby in this occasion, (R S) can obtain by R=(P-1) mod 5+1, S=(Q-6) mod 5+1 the table reference coordinates.

Then, and the table reference coordinates that above-mentioned steps S71 obtains (R, S), by the interpolation table of reference Figure 16, decision interpolating method (S74).

That is, in the interpolation table of Figure 16, indication is in that (R S) enlarges on the direction of record in the suitable coordinate range, generates the pixel data of the output pixel coordinates of record in this coordinate range with the table reference coordinates.In addition, in this occasion, indication A selects circuit to select the output of pattern match interpolating circuit 412.

On the other hand, in the interpolation table of Figure 16, (R, when S) suitable coordinate be blank, indication A selection circuit was selected the output of nearest interpolating circuit 411 with the table reference coordinates.

Then, according to the interpolating method of selecting, the following interpolation (S75) of carrying out.

(explanation of the operation of pattern match interpolating circuit 412)

Figure 18～Figure 20 is input imagery middle conductor width common pattern match map table that uses when being 1 pixel.In this map table, 3 * 3 pixel coverages of the left end of arrow are the patterns that is used to mate, the pattern of the pixel data of expression output in the scope on the right side of arrow, wherein expression output pixel coordinates in the thick line.

In addition, among Figure 18～Figure 20, the black picture element scope represents that line segment constitutes pixel, and the white pixel scope is represented line segment pixel on every side.In addition, a plurality of pixel coverages of diagonal line hatches represent that at least one pixel coverage is that line segment constitutes pixel, and the pixel coverage of horizontal line shade partly represents it can is that line segment constitutes pixel, also can be other pixels.

Pattern match interpolating circuit 412, according to the interpolating method of above-mentioned steps S74 decision, Figure 18～with reference to the pattern match map table of Figure 20.Then, from the input pattern of Figure 18～Figure 20, select the consistent input pattern of line segment information (the line segment information of 3 * 3 pixel coverages) that sends with above-mentioned characteristic detecting circuit 404, according to select with import the corresponding output pattern of pattern, generation the output pixel data.In addition, the black picture element of output pattern adopts line segment to constitute the average brightness MA of pixel among Figure 18～Figure 20, and white pixel adopts the average brightness MB of pixel on every side.

Concrete operations are as follows.

(1) occasion that enlarges on the horizontal direction

The occasion of Kuo Daing is in the horizontal direction carried out pattern match with reference to the pattern match map table of Figure 18.

At first, be in 3 * 3 pixels at center with the target pixel, it is the occasion of having only 1 pixel that line segment constitutes pixel, enlarges in the horizontal direction according to the pattern match map table 1. of Figure 18 (a).

With the target pixel is in 3 * 3 pixels at center, and line segment formation pixel is the occasion more than 2, the following pattern match of carrying out.

At first, have the occasion of the pattern that meets in 2. and 3. any one, adopt this pattern match map table.

On the other hand, in 2. 3. two occasions that condition all satisfies, at first, judge whether to meet shown in Figure 18 (b) one of them of exception pattern, if the occasion of the pattern that meets is arranged, then adopt this pattern match map table, if the occasion of the pattern that do not meet, then adopt 2. and 3. these two pattern match map tables.In this occasion, make the output image signal be 2. and 3. both sums.

In addition, suppose and can not occur making an exception and make an exception or the situation of the combination of exception and basic pattern.

(2) occasion that enlarges in vertical direction, identical with the occasion that enlarges on the horizontal direction of above-mentioned (1), carry out pattern match with reference to the pattern match map table of Figure 19.

(3) occasion that enlarges on both direction is carried out pattern match with reference to the pattern match map table of Figure 20.

With the target pixel is in 3 * 3 pixels at center, and it is the occasion of having only 1 pixel that line segment constitutes pixel, enlarges on both direction according to the pattern match map table 1. among Figure 20 (a).

With the target pixel is in 3 * 3 pixel coverages at center, and it is occasion more than 2 that line segment constitutes pixel, if (a) 2. 3. 4. 5. one of them have the pattern that meets, then adopt it to carry out pattern match.

On the other hand, the 2. 3. 4. occasion of the condition more than 2 in 5. in satisfying (a), at first, judge whether to meet shown in Figure 20 (b) one of them of exception pattern, if the occasion of the pattern that meets is arranged, then adopt this pattern match map table, if the occasion of the pattern that do not meet, then adopt the plural combination of 2. 3. 4. selecting 5. the pattern match map table from Figure 20 (a).In this occasion, the output image signal also is the pattern sum from 2. 3. 4. 5. selecting.

As the concrete example that carries out interpolation with above-mentioned pattern match, can investigate the image (identical) of input Figure 22 with Fig. 3 (b), and should table reference coordinates (R is S) with the consistent occasion of pattern shown in the input imagery of Figure 22.

For example, for the coordinate (2,3) of input imagery, owing to be equivalent to enlarge on both direction, thereby the map table that adopts Figure 20 carries out interpolation to 4 pixels (2,4), (3,4), (2,5), (3,5) in the output image of Figure 22.

The pixel data of input coordinate (23) be this as line segment constitute pixel and go up, right, all exist line segment to constitute pixel down.In this occasion, owing to be not suitable for the exception of Figure 20 (b), and be equivalent to make up 2. 3. 4. the occasion of Figure 20 (a), thereby in the output image after the conversion, (2,4), (3,4), (2,5) become line segment and constitute pixel (deceiving), pixel (in vain) around (3,5) have become.

For follow-up input coordinate (2,4), owing to be equivalent to enlarge in the horizontal direction, thereby can adopt the map table of Figure 18 that interpolation is carried out in two coordinates (2,6), (3,6) in the output pixel of Figure 22.

The pixel data of this coordinate (2,4) itself is that line segment constitutes in the pixel and upper and lower, the right side and all exists line segment to constitute pixel.In this occasion, because the Figure 18 (b) that is applicable to exception 3., thereby in the output image after the conversion, (2,6) become line segment and constitute pixel (deceiving), and (3,6) become pixel (in vain) on every side.

Thereby,, as shown in figure 22, form the output image of live width certain (1 pixel) by carrying out pattern match repeatedly.

According to the indication among the above-mentioned steps S74, A selects circuit to select one of them of pixel data after the conversion of the pixel data of nearest interpolating circuit 411 outputs and 412 outputs of pattern match interpolating circuit.

In addition, the pixel data of the pixel data of row of the 1st in the output image coordinate range of above-mentioned Figure 17 and the 2nd row and the 4th row and the 5th row is imported A simultaneously and is selected circuit 405.Thereby, in the 1st row of input, the 2nd row, in a single day the signal of the 2nd row stores line storage into, then behind the end of output of the 1st row, export the signal (the 4th row and the 5th row are too) of the 2nd row, thereby, reset according to the row preface with the picture intelligence after 405 pairs of conversion of A selection circuit, and send to B selection circuit 406.

But, except aforesaid way, also can in the preferential interpolating circuit 410 of live width, line storage be set, reset according to the row preface with the picture intelligence after 410 pairs of conversion of the preferential interpolating circuit of live width, and send to A selection circuit 405.

(B selects the operation of circuit 406)

If send result of determination (target pixel whether characteristic) from interpolation part detecting circuit 404, when then importing by the pixel data after center of gravity maintenance interpolating circuit 401 and preferential interpolating circuit 410 conversion of live width is parallel at every turn, B selects the result of determination of circuit 406 according to 404 inputs of characteristic detecting circuit, selects wherein side output.

This selection operation, shown in the process flow diagram of Fig. 6 of embodiment 1 explanation, according to result of determination from 404 inputs of characteristic detecting circuit, promptly, according to whether being characteristic, in the occasion of " non-characteristic ", the pixel data of selection center of gravity maintenance interpolating circuit 401 outputs (step S31 → S32), in the occasion of " characteristic ", and the pixel data of selection A selection circuit 405 outputs (step S31 → S33).

By above operation, target pixel before the conversion is the occasion of non-characteristic, selects the corresponding with it pixel data from center of gravity maintenance interpolating circuit 401, on the other hand, target pixel signal before the conversion is the occasion of characteristic, selects corresponding to it 410 pixel data.

As concrete example, the occasion of image shown in above-mentioned Fig. 3 (a) around the target pixel keeps interpolating circuit 401 to carry out conversion by center of gravity.

On the other hand, the occasion of image shown in above-mentioned Fig. 3 (b) carried out the number of pixels conversion by the preferential interpolating circuit 410 of live width around the target pixel.

(effect of present embodiment)

As mentioned above, by the number of pixels conversion process, for the place of interpolation one to one (coordinate of white among Figure 17), it is the occasion of 1 pixel width that the line segment in the former image constitutes pixel, also can keep 1 pixel width behind horizontal vertical both direction up conversion.In addition, the map table of Figure 18～Figure 20 is set for and is made that 1 pixel width also can be kept 1 pixel width after conversion in the former image, thereby pass through pattern match, in the place of having carried out interpolation (coordinate of shadow representation among Figure 17), the line segment of 1 pixel width in the former image can be kept live width.

Thereby, for character and picture,, can carry out fine rule and can not thicken in the either direction of level level, and the conversion that live width consistent is not destroyed, lines can not interrupt.

Particularly, in the present embodiment, owing to adopted the pattern match method when carrying out the preferential interpolation of live width, for many zones of reporting to the leadship after accomplishing a task of oblique line and line, for example the fine portion of line image also can be reproduced well by the number of pixels conversion, can keep its legibility for meticulous literal.

On the other hand, can not lose the conversion of natural image original appearance to natural image.In addition, the part beyond characteristic and the characteristic also can obtain smoother output image, even the input imagery that literal and image mix also can obtain the conversion image of good image quality.

(variation of present embodiment etc.)

Be that 1.2 times, vertical direction interconversion rate are that 1.4 times occasion is illustrated to the horizontal direction interconversion rate in the above-mentioned explanation, but as long as each interconversion rate of horizontal vertical is more than 1 times below 2 times, carry out the number of pixels conversion with the method identical, can obtain good number of pixels transformation results with said method.

In the above-mentioned explanation, to the live width that is detected by the characteristic detecting circuit is that the occasion of 1 pixel is illustrated, if possess the pieces that have much with more number of pixels and be object with reference to pattern, because the pixel of corresponding possible line segment is wide in the scope of broad, even then line segment constitutes the live width of pixel 2 occasions more than the pixel, also can by with reference to corresponding to each live width with reference to pattern, obtain good transformation results.

In addition, even the number of pixels interconversion rate is in the occasion more than 2 times, do not adopt input imagery coordinate and the man-to-man correspondence of output image coordinate, and the correspondence of employing one-to-many (integer after casting out below the decimal of number of pixels interconversion rate X), and increase the number of pixels of the piece be used for pattern match, also can obtain good transformation results.

(variation of the foregoing description 1～4 etc.)

* in the foregoing description, import the number of pixels interconversion rate by the operator, but also can adopt additive method, for example, also can judge the form of input image signal, the number of pixels of corresponding display device 20 is come automatic calculating pixel number interconversion rate.

* in the foregoing description 1～4, the situation that the number of pixels interconversion rate is applied to whole image is illustrated, but under the different situation of the number of pixels interconversion rate in the zone of each picture, also can implement the present invention equally.

* in the foregoing

description

1,3,4, pixel data to each input carries out characteristic whether judgement, with selecting circuit this result of determination is selected, but can whether be the judgement of part (natural image part) also, with selecting circuit to select according to this result of determination to each zone after the picture segmentation with feature of character and picture.

For example, in the feature line segment detecting circuit, in case store the picture intelligence of a picture, the judgement that then can whether have the feature of character and picture to each zone after the picture segmentation, in the occasion in the natural image zone that can differentiate picture according to the management information of the OS of computer 10 management and character and picture zone (specifically, when for example graphdriver carries out the number of pixels conversion, can know from management information whether each zone has the occasion of Word message), if this management information is sent to image-processing system, can whether be character and picture part or natural image part then to each picture segmentation regional determination.

* comprehensive transform method shown in transform method shown in the embodiment 1 and the embodiment 3.For example, can be comprehensively in the horizontal direction the transform method of embodiment 1 and the transform method of embodiment in vertical direction 3.

* also can be as embodiment 2 among the

embodiment

3,4, adopt the preferential interpolation method of live width at interconversion rate X near the occasion of integer.

* in the foregoing description 1～4, illustrated and handled between each pixel data of input in turn mended and the example of output, same, also can carry out interpolation and handle the area unit after the picture segmentation.

* the processing in the image-processing system of explanation in the foregoing description 1～4 (processing shown in Fig. 2,4,5,6,9,11,12,21 the process flow diagram) can be realized by software, goes for computing machine.

Utilize possibility on the industry

The present invention be applicable to be positioned at from image signal source to PDP, the array display device of LCD etc. Image-processing system on the path of transmission picture intelligence. Or be applicable to array display device In built-in image-processing system.

Claims

1. A kind of image processing method, the original image that input matrix shows is used, carries out pixel number transformation to at least one designated direction in row direction and column direction with predetermined pixel number transformation rate, and it comprises the following steps:

The edge detection step detects the edge constituent pixels contained in the input original image;

The output adjustment step is to multiply the number of continuous edge constituent pixels detected by the edge detection step by the integer uniquely determined based on the pixel number transformation rate, and adjust the transformed edge constituent pixels to be in the Continuous in the specified direction.

2. The image processing method as claimed in claim 1, characterized in that:

In the edge detection step,

A pixel having a gradation difference greater than or equal to a predetermined value with respect to adjacent pixels in the specified direction in the original image is detected as an edge constituting pixel.

3. The image processing method as claimed in claim 1, characterized in that:

In the edge detection step,

The spatial frequency is measured for each portion of the original image, and edge constituent pixels are detected based on the measurement result.

4. image processing method as claimed in claim 1, is characterized in that:

The integer uniquely determined based on the conversion rate of the number of pixels,

When the pixel number conversion rate X is 1 when 0<X<0.5, and is N when N-0.5≤X<N+0.5, N is an integer greater than 1.

5. image processing method as claimed in claim 1, is characterized in that:

In the output adjustment step,

For the edge constituting pixels detected by the edge detection step, the output is based on the pixels after the integer transformation uniquely determined based on the transformation rate of the number of pixels,

For pixels other than the edge constituting pixels detected in the edge detection step, the number of converted pixels outputted is changed pixel by pixel.

6. The image processing method as claimed in claim 1, characterized in that:

For pixels other than the edge constituting pixels detected in the edge detection step,

An integer is selected from an integer uniquely determined based on the conversion rate of the number of pixels and an integer obtained by adding or subtracting an integer within a prescribed range to the integer, and the number of pixels corresponding to the integer is output.

7. An image processing method, inputting the original image used for matrix display, carrying out the pixel number conversion to at least one designated direction in the row direction and the column direction with a predetermined pixel number conversion rate, it is characterized in that it selects the first One of the 1st interpolation method and the 2nd interpolation method performs pixel number conversion, wherein:

A first interpolation method for outputting a converted image corresponding to the input image at a certain magnification determined based on the conversion rate of the number of pixels;

The second interpolation method is to output the transformed pixels corresponding to the input pixels by performing pattern matching on the pattern of the line segments constituting the pixels contained in the input original image and the predetermined conversion table

8. A kind of image processing method, the original image that input matrix shows is used, carries out pixel number transformation to at least one specified direction in row direction and column direction with predetermined pixel number transformation rate, it is characterized in that comprising:

The first interpolation step is to output a converted image corresponding to the input image at a certain magnification based on the conversion rate of the number of pixels;

The 2nd interpolation step, by carrying out pattern matching to the pattern that the line segment constituting pixel that input original image comprises and predetermined transformation table, output the pixel after the transformation corresponding to input pixel;

A selection step selects the output of one of the first interpolation step and the second interpolation step.

9. image processing method as claimed in claim 8, is characterized in that: have the feature determination step of whether each region of original image is judged one by one feature part,

In the selection step,

In the feature determination step,

For the image part that is determined as a feature part, select the first pixel number conversion method, so that the line segments with the same line width in the original image of the input also maintain the same line width after conversion;

For the image part judged to be a non-characteristic part, the second pixel number transformation method is selected so that the center of gravity position before and after transformation remains unchanged.

10. The image processing method as claimed in claim 9, characterized in that:

In the feature determination step,

According to the information of the line segments contained in the original image, each part of the original image is judged one by one whether it is a characteristic part or not.

11. The image processing method as claimed in claim 9, characterized in that:

In the feature determination step,

When the target pixel in the input image contains line segment pixels below the specified line width in a certain area centered on the target pixel, and the variation coefficients of the luminance value of the line segment pixel group and the luminance value of the surrounding pixel group are respectively within the specified value or less, and the difference between the average grayscale value of the pixel group constituting the line segment and the average grayscale value of the surrounding pixel group is greater than or equal to a specified value, it is determined as a characteristic part.

12. The image processing method according to claim 11, characterized in that:

The specified line width is 1 pixel.

13. The image processing method as claimed in claim 9, characterized in that:

In the feature determination step,

The determination of whether it is a characteristic part or not is carried out one by one for each area after one screen division of the original image.

14. An image processing method, one of the following methods is selected to transform the number of pixels:

The first pixel number conversion method makes the line segments with the same line width in the input original image also maintain the same line width after transformation;

The second pixel number conversion method keeps the center of gravity position before and after the conversion unchanged,

include:

The characteristic judgment step is to carry out the judgment of whether the characteristic part is carried out one by one to each region of the original image;

Selecting step, for the image portion determined as a feature portion by the feature determination step, select the first pixel number conversion method, and for the image portion determined as a non-characteristic portion, select the second pixel number conversion method ,

In the feature determination step,

15. The image processing method as claimed in claim 14, characterized in that:

The specified line width is 1 pixel.

16. An image processing method, select one of the following methods to convert the number of pixels:

include:

In the selection step,

Determine whether the value of the conversion rate of the number of pixels is close to an integer,

When it is judged to be close to an integer, the first pixel number conversion method is selected, and when it is judged not to be close to an integer, the second pixel number conversion method is selected.

17. An image processing method, select one of the following methods to transform the number of pixels:

The first pixel number conversion method makes the line segments with the same line width in the input original image also maintain the same line width after conversion;

include:

Selecting step, for the image portion determined as a feature portion by the feature determination step, select the first pixel number conversion method, and for the image portion determined as a non-characteristic portion, select the second pixel number conversion method , the selection step,

When the difference between the value of the pixel number conversion rate and an integer close to it is less than a specified value, the first pixel number conversion method is selected, and when it is greater than the specified value, the second pixel number conversion method is selected.

18. An image processing device, inputting the original image used for matrix display, carrying out the pixel number conversion to at least one designated direction in the row direction and the column direction with a predetermined pixel number conversion rate, and it includes the following parts:

The edge detection part detects the edge constituent pixels contained in the input original image;

The output adjustment part, for the edge constituting pixels detected by the edge detection part, in the specified direction, transforms After that, the edge constituting pixels are adjusted to be continuous.

19. The image processing device according to claim 18, characterized in that:

In the edge detection section,

20. The image processing device according to claim 18, characterized in that:

In the edge detection section,

21. The image processing device according to claim 18, characterized in that:

22. The image processing device according to claim 18, characterized in that:

The output adjustment section,

For the edge constituting pixels detected by the edge detection part, the pixels after integer transformation uniquely determined based on the transformation rate of the number of pixels are output,

For pixels other than the edge constituting pixels detected by the edge detecting section, the number of converted pixels outputted is changed pixel by pixel.

23. The image processing device according to claim 22, characterized in that:

For pixels other than the edge constituting pixels detected by the edge detection section,

24. An image processing device that inputs an original image for matrix display, and performs pixel number conversion on at least one designated direction in the row direction and the column direction with a predetermined pixel number conversion rate, and is characterized in that it selects the first One of the 1st interpolation method and the 2nd interpolation method performs pixel number conversion, wherein:

The second interpolation method is to output the transformed pixel corresponding to the input pixel by pattern matching the pattern of the line segment constituting the pixel contained in the input original image and the predetermined conversion table

25. An image processing device, inputting the original image used for matrix display, carrying out the pixel number conversion to at least one designated direction in the row direction and the column direction with a predetermined pixel number conversion rate, characterized in that it comprises:

The first interpolation part outputs a converted image corresponding to the input image at a certain magnification determined based on the conversion rate of the number of pixels;

The 2nd interpolation part, by carrying out pattern matching with the pattern of the line segment constituting the pixel that the input original image comprises and predetermined transformation table, output the pixel after the transformation corresponding to input pixel;

A selection section selects an output from one of the first interpolation section and the second interpolation section.

26. An image processing device that selects one of the following methods for pixel number conversion:

include:

The feature determination part is to determine whether each area of the original image is a feature part one by one;

Selecting part, for the image part judged to be a characteristic part in the characteristic judging step, select the first pixel number conversion method, for the image part judged as a non-characteristic part by the characteristic judging part, select the first Two-pixel number transformation method,

The characteristic determination part,

27. An image processing device that selects one of the following methods for pixel number conversion:

include:

The specified line width is 1 pixel.

28. An image processing device that selects one of the following methods for pixel number conversion:

include:

In the selection section,

29. An image processing device, one of the following methods is selected for pixel number conversion:

include:

In the selection section,